Watchpoint killing memory

Any thoughts?  Anyone?  Bueller?  TI?

Could you please provide more details about the code that is being executed?

also, is the following the steps to reproduce ?

1. execute your code

2. halt the cpu.

3. inspected the value @ 0x01C2000. 

4. apply watchpoint on write access @ 0x64004F94 

5. inspect value @ 0x01C2000, value does not match the one seen in step 3 ?

Also, what version of CCS are you using?

visu_l,  thank you for looking at my problem.

The program we are running doesn't appear to be important, we initialize the timer:

  time_clock_regs.TIM12 = 0x00000000u;
  time_clock_regs.TIM34 = 0x00000000u;
  time_clock_regs.TGCR  = TIM12RS_NOT_IN_RESET | TIM34RS_NOT_IN_RESET;
  time_clock_regs.PRD12 = 0x00000000u;
  time_clock_regs.PRD34 = 0x00000000u;
  time_clock_regs.TCR   = TIM12_RUNS_CONTINUOUSLY | TIM34_RUNS_CONTINUOUSLY;

then check the timer value in a loop:

  current.LSB = time_clock_regs.TIM12;
  current.MSB = time_clock_regs.TIM34;

We can step through code in CCSv5.1.0.09000 and see TIM12 changing all by itself when we look at the memory from 0x01C20000.

As soon as we add a watchpoint, all data in the memory window from 0x01C200000 is zero.  The value of TIM12 is always zero, even if we remove the watchpoint, restart program.  

The only way to reset seems to be to close CCS and restart.

I've created a simple example that demonstrates the problem.

The attached zip file contains a simple main.c and a test.js DSS script that drives it.  If you have CCSv5 installed in c:\ti you can run test.bat to build and run the example.

8420.watchpoint_error.zip

I have tested this code  CCSv5.1 (upgrade pack 7, now listed as version 5.1.0.09000) with the same results.  Has any progress been made on this issue?